From building softsynths to composing in first-gen DAWs. From LAN-party Quake player to global e-sports team leader. From a weekend job selling kitchens to 3D and VR modelling interiors.
Experimentation is how I learn. Going deep is where I find the parts that — if changed — move everything.
This site is old hobbies and new hobbies, revisited with AI on the bench. Some of what comes out translates into business. I work in tech and media by day. Homelab of agents I tinker with at night. Almost nothing here took longer than an evening to start — that's rather the point.
This site is AI-assisted and human-directed. Every image, line, and clip was AI-generated, then selected, edited, and approved by me. Personal views, not affiliated with my employer.
Side-line learnings worth sharing — from my own AI experiments, the day job, and the networks I sit in. Notes on creative AI in music, video, and games. And a log of the technical projects running in the homelab.
I experiment with AI a lot — mostly to keep up, and mostly in the corners I find genuinely fun: where it's exciting, and every so often where it's actually practical. Some of it turns out useful, some of it is just play, and all of it is made in a light spirit. None of it is finished. All of it is up for discussion. Drop a note if any of it is useful, broken, or worth arguing about.
The ones I keep coming back to in real rooms with real P&Ls. Written for the CMO or CEO with thirteen minutes and one honest question.
Senior leaders are treating a generational productivity tool as a way to chop margin. The ATM, the spreadsheet, and a 2026 PwC study all tell the same story: history punishes the cost-cutters. Notes on the inversion — and the boardroom move that follows.
Read the full piece
When marketing numbers wobble, the corporate reflex is to tear down the kitchen — new agency, new stack, new eighteen-month RFP. The empirical data says that is the most expensive mistake in the playbook. Notes on the three variables you actually need to tune.
Read the full piece
Org change runs at twelve to eighteen months. The model that lands in production next quarter is six weeks away. You cannot wait for the reorg to finish before integrating AI — you have to build the team so an agentic teammate can sit down on day one. Notes on what that actually looks like.
Read the full piece
Forty unpaid volunteers, no salaries, no contracts, no HR — and a kill rate most companies would envy. A top World of Warcraft guild had to make the systems carry everything: incentives that keep the floor high, blameless logs, and a bar that belonging enforced. Notes on what the org chart can steal from the raid.
Read the full piece
Everyone rents an AI chief of staff that forgets them at midnight. Notes on the alternative: small skills that read and write plain markdown files, so the intelligence accumulates in files you own — designed with a frontier model, run on a modest one.
Read the full piece
Hold people accountable for the what — and let them own the how. A new-job playbook with a playable exhibit: carry water, adopt the memos, and watch the colander become the job.
Read the full pieceI run these to find out what the tools can actually do. Each one is half-finished on purpose. The asset isn't the output — it's the prompt set that produced it.
TODAY'S RENDER
THE PROMPT
A dark, cinematic editorial photograph in an industrial-meets-luxury register -- Rotterdam warehouse meets designer atelier. a tailored wool overcoat hanging on a steel coat rack; blue hour, wet asphalt sheen reflecting the last light; long-exposure motion blur around a tripod-still subject. The subject emerges from a gradient-to-black ground: deep charcoal #0a0a0c fading to soft graphite #1e1e22. Photorealistic photography, not painted, not illustrated, not CGI. No people, no faces, no text, no captions, no logos, no watermarks.
Model: gemini-3.1-flash-image
Built locally, broken locally, documented in the order I figured them out. Not a portfolio — a log of what runs on top of which other piece.
Voice-first deep-research agent. Multi-phase pipeline (query analysis → study planning → iterative source-grounded research → claim validation → synthesis) that reads the result back aloud. Google ADK + Gemini + ElevenLabs on Cloud Run.
Voice-first family assistant for the car. Calendar, meal planning + Picnic ordering, and a deep-research / debate system. ElevenLabs voice in, WhatsApp out for approvals.
AI content factory for LinkedIn. Four-room workflow (ingredients → AI writing → visual generation → publish), twelve writing personas, six visual archetypes. .NET 8 + React 19 + Gemini.
Dutch AI bedtime-story generator. Build a fantasy world with characters, configure age + arc + learning goals, get an illustrated multi-chapter story with parent reading aid and printable booklets. Public portfolio version + private family deployment.
Multi-agent debate arena. Pro + Con researchers, Pro + Con debaters, a moderator, a judge. Four battle modes: debate, rap battle, roast, pitch-off. Three personality knobs.
Interactive Jiskefet-inspired turn-based simulator. Survive 30 turns of exponential AI upgrades and reach a $100B valuation.
A segmented home network (open-source firewall, several VLANs) plus a self-learning custom climate controller and a BigQuery-backed thermal model that arbitrages against dynamic Dutch energy prices. The infrastructure layer everything else runs on — kept deliberately high-level.
Agentic creative-agency tooling for YouTube. 10 Cloud Run services + Next.js cockpit + Composer DAG. Static brand assets in, 16:9 + 9:16 variants out, compliance-judged and Google Ads ready. Built so one CSD runs the full workflow.
Bios, photos, topics — all on one page. Print as a PDF straight from here if that's what your workflow needs.
Note — this is a personal site and the kit below is for media convenience. All assets and copy are AI-assisted, human-directed. Personal views, not affiliated with my employer. Portraits below are AI-generated; label them appropriately when used in print or press.
Rutger Tuit is a musician and tinkerer who ended up working where technology meets human creativity — looking for the strategic reasons behind the technical detail, and helping the people who have to act on it. He picks up a new field well enough to spot where it's about to bump into the next one, and tries to flag that early. Right now that's in marketing: he's Director, Specialists & Partners, Google Benelux, and Head of YouTube, and sits on Google's Northern European Sales Leadership Team. The way he likes to work — high trust, teams that are loosely coupled but agree on the goal, plenty of room to experiment — isn't really specific to marketing. This site is a personal project; the views here are his own.
Six AI-generated portraits (1440×1920 PNG) plus the three bios above, bundled as one zip. Each image is synthetic — not a photograph. Label as such if used in print or editorial.
Plus 24 interviews and conversations — see Articles & Interviews Elsewhere.
For public speaking I only participate in approved professional contexts related to my role — topics anchored to Google products and the YouTube platform. This site itself is personal; outside-the-day-job engagements run through that approval process, not through this kit.